Skip to main content
Create a path that the agent can trigger autonomously when needed.
Agentic Tool nodes define when the AI should call a tool and what inputs it must provide. It is a gateway to the nodes that follow that determine how the agent behaves afterward.

​
Main Settings

​
Tool Name (required)

The name the AI will reference when deciding whether to call this tool.
Choose a name that clearly communicates its purpose (e.g., lookup_order, create_ticket).

​
When Should AI Use This Tool?

Provide clear guidance describing:
  • When the tool should be triggered (conversational cues)
    Example: “Use this when the user asks about their account balance.”
  • Why the tool should be called (functional purpose)
    Example: “Use this to retrieve inventory information.”
This text directly influences the agent’s decision-making.

​
AI Inputs Return Mode

Specifies how the AI should return the values it generates for this tool. Options:
  • Single Object — AI returns a single object.
  • List of Objects — AI returns an array of objects, useful when the tool may need to operate on multiple items.

​
AI Tool Inputs (Parameters)

Define the parameters the AI should generate and provide when invoking this tool.
Each parameter becomes a variable accessible within the tool’s workflow path. Mark inputs as required if the tool cannot function without them, or optional if they are not always needed. You may add multiple inputs; each contains the following fields:

​
Variable Name (for AI to use)

The internal name used to store this parameter (must start with a letter and use only letters, numbers, and underscores).
Examples: customer_name, quantity, is_member.

​
Input Type

What type of value the AI should generate. Types include:
  • Text (String)
  • Number
  • Integer
  • True/False (Boolean)

​
Input Format

Whether the parameter is:
  • Single Value
  • List of Values

​
Description (for the AI)

Explain what the input represents so the AI knows what to produce.
Examples:
  • “The customer’s full name”
  • “Number of items in the order”
  • “Whether the user accepted the terms”

​
Required Input?

If enabled, the AI must provide this value for the tool to execute.
If disabled, the tool can run even if this parameter is missing or null.

​
After Tool Execution

Controls how the agent behaves after the tool finishes running. Options:
  • Silent (only respond if ‘Tool Response’ node added)
    The agent returns to listening without speaking. To provide spoken output, add a Tool Response node.
  • Always have agent speak about what happened
    The agent automatically explains the tool result, even without a Tool Response node.